Matthew C. Nowlin
University of Oklahoma
Over the last two decades many alternate theories of the policy process have been developed. This essay covers recent scholarship (from 2008 to 2010) regarding the major policy process theories. In addition, several recent trends in research are discussed including; the use of narrative in policy theory, issues that cross multiple subsystems, bureaucracy in the policy process, and synthesizing multiple theories and frameworks.
Policy scholarship is divided between knowledge in the policy process and knowledge of the policy process (James & Jorgensen, 2009; Lasswell, 1971; Weimer, 2008). Knowledge in the policy process largely refers to knowledge produced through analysis and evaluation (James & Jorgensen, 2009), whereas knowledge of the policy process is "focused on the how and why of policymaking" (Smith & Larimer, 2009, p. 6). Early ways of understanding the policy process included Lowi's typologies (Lowi, 1972) and, most notably, the "stages" heuristic (see deLeon, 1999 for a review). Scholars became dissatisfied with the stages heuristic and called for "better" theories of the policy process (Nakamura, 1987; Sabatier, 1991). In particular, Sabatier (1991) noted that the stages heuristic had "outlived its usefulness and must be replaced" (Sabatier, 1991, p. 147). One of the major criticisms of the stages heuristic was that it did not contain any causal mechanisms; therefore it was not a scientific theory (Sabatier, 2007, p. 7). As dissatisfaction with Lowi's typologies and the stages heuristic grew, a number of alternative theories of the policy process began to proliferate (see Sabatier, 2007). This essay offers a brief assessment of the current state of policy process research and highlights several emerging trends.
This essay lays out the current policy process theories and frameworks and discusses recently published peer-reviewed research. The frameworks and theories to be discussed are drawn from Paul Sabatier's (2007) Theories of the Policy Process.1 Frameworks and theories included in this essay are the Institutional Analysis and Development Framework (IAD), Multiple Streams (MS), the Advocacy Coalition Framework (ACF), Policy Diffusion, Punctuated-Equilibrium (PE), and Social Construction and Policy Design. Recent research (largely from 2008 to 2010) was chosen from a search of major political science, public policy, and public administration journals including; American Political Science Review, American Journal of Political Science, Journal of Politics, Political Research Quarterly, Policy Studies Journal, Journal of Policy Analysis and Management, Review of Policy Research, Journal of Public Policy, Journal of Public Administration Research and Theory, and Public Administration Review. In addition, web searches using each framework or theory as search terms were conducted using both Google Scholar and ISI Web of Science. Research was chosen based on (i) empirical application, (ii) innovative methodology, and (iii) theoretical contribution and/or extension.2 The major factor for selection was theoretical insights and expansions of the theory or framework. Finally, policy area-specific research or journals are, for the most part, not examined.
The first section of this essay discusses recent research related to each of the major theories or frameworks. I give brief summaries of exemplar works that provide theoretical contributions (themes tying together several papers are discussed where possible) and I then provide a summary paragraph tying these recent works back to the theory or framework. The second section discusses emerging trends and growing areas of research. These trends include the Narrative Policy Framework, recent work on subsystems, trans-subsystems, and policy regimes; the role of the bureaucracy in the policy process, and ways to synthesize the various theories and frameworks.
The Institutional Analysis and Development (IAD) framework grew from the institutional rational choice (IRC) literature (Ostrom, 2007b). The IAD framework began as a book chapter (Kiser & Ostrom, 1982) examining the impacts of institution arrangements on human behavior, and has since grown into a fully developed research program. Its main proponent, Elinor Ostrom, was recently awarded a Nobel Prize in Economics.3 The IAD framework focuses on institutional arrangements in collective action settings and largely deals with common pool resources4 (see Ostrom, 2007b for a review). The IAD framework encompasses a theory of common pool resources, but for the purposes of this essay I focus on recent research based on the aspects of the framework that deal with institutions.5
Using collaborative watershed partnerships, Hardy and Koontz(2009) examine the type of groups involved, the rules these groups implement across the various levels of action (constitutional, collective choice, operational), and the subsequent impacts on watershed management. The three types of groups that they analyze are "government centered" (government agency based), "citizen centered", and "mixed" (both government agencies and citizen groups). Hardy and Koontz(2009) find that these groups tended to construct "rules-in-use" that varied across each level. The authors conclude that government and "mixed" groups tended to be similar in the type of rules they implemented (largely more formal and embedded in federal and state policy), whereas citizen groups tended to implement less formal agreements. The IAD framework identifies various institutional levels of rule-making authority, and Hardy and Koontz(2009) build on this insight by examining the way that different types of collectives can construct varying types of rules. This work adds greater understanding to the importance of citizen and government groups, how these groups create rules, and the subsequent impacts of those varying rules on policy outcomes.
One of the major insights offered by the IAD framework is that individuals can create self-governing institutions that mitigate conflicts (Ostrom, 2007b). Using 14 interstate water compacts negotiated by states in the western U.S., Schlager and Heikkila (2009) empirically test the ability of these compacts to reduce conflict. Specially, the authors compared the ability of water compacts to reduce conflict versus other institutions such as courts, legislatures, and government (federal and state) agencies. The authors found that, of the 23 resolved conflicts studied, five were resolved by government agencies or legislatures, eight were resolved by compact commissions, and ten were resolved through courts (Schlager & Heikkila, 2009, p. 382). Compact commissions were most likely to alter operational-level rules, while courts were most likely to alter collective choice and/or constitutional-level rules. The IAD emphasizes self-governing institutions, and this article has provided important empirical support for the role of interstate compacts in resolving conflicts. In addition, by analyzing the type of rules that are typically altered through different venues, the authors are able to relate various policymaking venues with the IAD. Overall, interstate compacts seemed to be most effective at mitigating conflicts at the operational-level, whereas the courts were more effective at reducing constitutional-level conflicts.
Within the IAD, institutions are defined as the "shared concepts used by humans in repetitive situations organized by rules, norms, and strategies" (Ostrom, 2007b, p. 23). Based on this definition, Crawford and Ostrom (1995) laid out a grammar of institutions. This institutional grammar can guide analysts in the identification of "institutional statements" which contain the strategies, rules, and norms that constitute a given institution. Recent work by Basurto et al. (2009) has developed a coding scheme for legislation, and/or other policy documents, based on this institutional grammar. Within their coding et al. define rules as prescriptions enforced by sanctions, norms as prescriptions enforced by inducements, and strategies as "regularized plans" made given the present set of incentives and the expected behavior of others (2009, 15). Basurto et al. (2009) then proceed to apply their coding strategy to two pieces of legislation; a federal transportation bill and a bill from the state of Georgia regarding abortion. They found that each bill contained many more statements of norms then of rules or strategies. In fact the transportation bill seemed to contain only norm statements (it contained 128 in all). The abortion bill contained 110 norm statements, 3 strategy statements, and 4 rule statements. This work offers a way to empirically examine the types of institutional statements that the IAD framework would expect. This coding scheme could be useful to scholars examining the type(s) of institutions (rules, norms, strategies) embedded in policy designs, and the possible impacts of these institutions on policy outcomes.
In two recent papers, Ostrom (2007, 2009) offers an expansion of the IAD framework to include linked social-ecological (e.g., human-environment) systems (SESs). The focus of this expanded framework is on making SESs sustainable by integrating understandings of the ecological system in question with the governance structure overseeing that system. The framework includes the following attributes "(i) a resource system (e.g., fishery, lake, grazing area) (ii) the resource units generated by that system (e.g., fish, water, fodder) (iii) the users of that system, and (iv) the governance system" (Ostrom, 2007, p. 15182). Of interest is the outcomes produce by (i) the interaction between these various attributes and (ii) the interaction between higher-level attributes such as the larger socioeconomic, political, and ecological systems. This expansion of the IAD offers an interdisciplinary framework that includes both social and natural sciences, and is focused on the sustainability of common pool resources.
The IAD framework is the only major policy theory or framework to be based on institutions. The work summarized here expands our understanding of institutions within the IAD. Hardy and Koontz (2009) examined how institutional design (rules) varied both by groups and levels of action. Schlager and Heikkila (2009) identified how self-governing institutions (interstate water compacts) were able to mitigate conflicts even when other government policymaking venues were included. One issue with the broader IRC literature, including the IAD framework, is that it seldom takes multiple institutions into account (Lubell, Henry, & McCoy, 2010). The work by Schlager and Heikkila (2009) incorporated multiple policymaking institutions and found support for the IAD framework. In addition, other work has argued that collaborative institutions are intermixed in an "ecology of games." This ecology contains several overlapping and multijurisdictional institutions that impact policy outcomes (Lubell et al.). It is expected that future work will continue to examine the impacts of multiple, inter-connected institutions. Basurto et al. (2009) developed a coding scheme for scholars to analyze institutional and policy designs that scholars should find extremely useful. Finally, Ostrom (2007, 2009) expands the IAD framework to include institutions embedded within broader social, political, and ecological systems.
Multiple Streams (MS) was developed by John Kingdon (1984) in his book Agendas, Alternatives and Public Policies. It posits that there are three separate and independent streams related to policy making; the problem stream, the politics stream, and the policy stream. Elements of the problem stream include the issues that "policy makers and citizens want addressed" (Zahariadis, 2007, p. 70). The politics stream consists of the national political environment, which can include public opinion and the partisan control of policymaking institutions. Finally, the policy stream consists of ideas and solutions, developed by experts and policy specialists, waiting to be implemented. Policy change occurs when a "window" of opportunity opens and a policy entrepreneur merges the three streams by applying an idea from the policy stream to an issue in the problem stream at a time when the problem/solution coupling is acceptable within the political stream.
Within MS, advocates and experts in the policy stream have ideas for policies, and monitor the problem stream for a condition amiable to their solution. Recent work has dubbed this "problem surfing" (Boscarino, 2009). According to Boscarino, problem surfing is defined as advocacy groups attaching "their preferred policy solution to whatever problem(s) [is] salient at the time" (2009, p. 421). In the area of forestry policy, Boscarino (2009) found that two interest groups, the Wilderness Society and the Sierra Club, tended to connect "their solution of sustainable forestry to different policy problems at different times" (p. 426). In addition, Boscarino (2009) found that both groups were sensitive to broader, high salience6 issues including the economy, climate change, and water quality. These groups would alter their message to include reference to these higher salience issues. One of the criticisms of MS is that it does not offer clear hypotheses that are falsifiable (Sabatier, 2007). However, Boscarino (2009) is able to empirically tests hypotheses generated from MS and finds results consistent with what MS would expect.
One of the major assumptions of MS is that each stream operates independently from the others. Using participation in school violence prevention, Robinson and Eller (2010) examine the assumption of separate problem and policy streams. The separate streams assumed in MS imply that participation in one stream limits participation in another. Using a survey of school superintendents, Robinson and Eller (2010) find that, counter to this assumption, many individuals and organizations participate in both problem identification (the problem stream) and policy proposals (the policy stream). This finding provides some empirical evidence that calls into question one of the major assumptions of MS.7
Building on MS, recent work has put forth a revised multiple streams model (Ness, 2010; Ness & Mistretta, 2009). One revision involves adding institutional factors, termed policy milieu. The policy milieu includes such institutions as state government structures, and state higher education governance structures (the revised multiple streams model is tested in the higher education policy domain). In addition, the revised model expands the policy stream into a "policy field" that contains the politics and problem stream. The assumption behind this expansion of the policy stream is that, "policy trends and information are present throughout the policy process" (Ness & Mistretta, 2009, p. 492). The policy trends within the policy field are comparable to adoption trends in the policy diffusion literature, while the policy information is comparable to information sharing between and across coalitions in the ACF (Ness & Mistretta, 2009). Policy entrepreneurs, like in the original MS theory, seek to merge the streams (in this case the politics and problem streams) when a window of opportunity is present.
In the revised model of MS, the focus is shifted from agenda setting to one of policy design and formulation. Entrepreneurs, located in the policy field, seek to merge the streams in order to ensure their preferred policy design is implemented. Using a comparative case study approach, Ness and Mistretta (2009) find that the revised MS model accounts for the policy design differences between two neighboring states (Tennessee and North Carolina) with regard to use of a lottery to fund higher education. Specifically, the authors found that "state government structure, related intra-state policy trends, policy entrepreneurs, and the timing of policy windows mattered most" (Ness & Mistretta, 2009, p. 509). Additional research found the revised MS model was able to provide insight into policy design choices regarding need-based vs. merit-based college aid in New Mexico, Tennessee, and West Virginia (Ness, 2010).
As noted, MS is often criticized for not producing empirically testable hypotheses. However, each of the articles discussed provided empirical applications of MS. Boscarino (2009) found some support for MS by empirically testing how groups surf the problem stream for salient issues to attach to or reframe their solutions. Robinson and Eller (2010) however did not find support for the independent stream assumption of MS. Finally, a revised MS model moved beyond agenda setting by offering empirical insight into policy design choices (Ness, 2010; Ness & Mistretta, 2009). The revised model of MS seems to have only been tested within the higher education domain; further work should empirically test this model across various policy domains.
The Advocacy Coalition Framework (ACF) was initially devised by Paul Sabatier and Hank Jenkins-Smith (Jenkins-Smith, 1990; Sabatier, 1987, 1988; Sabatier & Jenkins-Smith, 1988), and was later expanded and clarified by Sabatier and Jenkins-Smith (1993, 1999) and Sabatier and Weible (2007). The focus of the ACF is on policy learning and policy change within a policy subsystem. Policy change was initially thought to occur as a result of policy learning or external shocks. External shocks include public opinion, changes in governing coalitions, and outputs from other subsystems (Sabatier & Jenkins-Smith, 1999). More recent iterations of the ACF have added internal (subsystem) shocks, and negotiated agreements between coalitions as factors influencing policy change (Sabatier & Weible, 2007).
Central to advocacy coalitions are shared policy core beliefs that are shaped by more abstract core beliefs. The ACF argues that shared beliefs result in coalitions that are homogenous (with regard to beliefs and patterns of coordination) and stable overtime. Weible, Sabatier, and McQueen (2009) examined over 80 applications of the ACF and found that coalitions are largely stable, particularly among "principal" members. However, coalition defections occur and coalitions are not necessarily homogenous in their beliefs. Some coalition members can vary in their policy core and secondary aspect beliefs and sub-coalitions may also exist.
Coalition homogeneity may be undermined by the political and/or self-interests of coalition members (Nohrstedt, 2010; Szarka, 2010). A long line of research in the ACF has examined the possible importance of interests with regard to coalition homogeneity and stability (see Szarka, 2010, pp. 838–40 for a review). More recently, Nohrstedt (2010) finds that interests played a large role in nuclear policy making in Sweden. Political parties often made policy decisions based on strategic political concerns rather than normative beliefs. Another important aspect of coalition stability may be trust (Lubell, 2007). Lubell (2007) finds that shared policy core beliefs are predictive of the level of trust that coalition member's exhibit. Other work however, has noted that narrow coalitions based on normative beliefs may not be broad-based enough to encourage substantial policy change (Ansell, Reckhow, & Kelly, 2009). In sum, coalitions have been demonstrated as stable, however political or material interest may undermine their homogeneity. On the other hand, homogeneity may weaken the ability of a coalition to bring about policy change.
In addition to coalitions, the ACF has traditionally been concerned with sub-system dynamics. However, work by Jones and Jenkins-Smith (2009) argues that ACF scholars should examine more macro-level, trans-subsystem features of the policymaking system. These macro-level features include clusters of linked sub-systems, public opinion, and policymaking venues. These features constitute the policy topography in which policy actors operate. Jones and Jenkins-Smith (2009) argue that public opinion, underutilized in ACF applications, is the foundation of the policy topography. They contend that shifts in public opinion can act as an exogenous shock and shift the policy topography and/or act as an endogenous shock, causing shifts within a particular subsystem.
Apart from public opinion, Jones and Jenkins-Smith (2009) offer two other types of possible shocks; salience disruptions and dimension-shifts. Focusing events and events within proximate subsystems can act as exogenous shocks to a subsystem causing a salience disruption. The 9/11 terrorist attacks are offered as an example of a salience disruption. A type of internal shock mentioned by Jones and Jenkins-Smith (2009) are policy dimension-shifts. Policy dimension-shifts occur when strategic policy entrepreneurs import arguments from another proximate linked subsystem. The scientific consensus regarding anthropogenic climate change is given as an example of an issue that could possibly be used to cause a dimension-shift within one of the subsystems linked to climate change (e.g., coal energy, air quality and pollution, and nuclear energy). The paper by Jones and Jenkins-Smith (2009) offer two important expansions to the ACF; the first is the role of public opinion as both a constraint on coalition strategy and a resource, and the second is the expansion of the ACF beyond subsystems (traditionally assumed to be independent and self-contained) to a policy topography model.
Recent work summarizing over 80 applications of the ACF (Weible et al., 2009) demonstrated that the ACF has developed into a strong research program; with a growing number of applications outside the United States and across several policy areas. The work discussed above examined one of the ACF's key assumptions; the stability and homogeneity of advocacy coalitions. Overall, coalitions have been found to be relatively stable overtime, but not consistently homogenous. Stability and homogeneity may be undermined by the interest considerations of some coalition members. Weible et al. note that future work should differentiate between principal and auxiliary coalition members (2009, 130). The conjecture would be that principal coalition members are more likely to be both stable and homogenous, whereas auxiliary members may demonstrate stability but are less likely to be homogenous. Apart from coalitions, Jones and Jenkins-Smith (2009) argue that ACF scholars should examine the trans-subsystem dynamics of the policy topography model. This model offers insights based on public opinion and strategic policy entrepreneurs that can be incorporated into a fuller understanding of policy learning and change.
Policy diffusion research tracks how similar policy innovations are adopted across states in the American context, or across countries in comparative contexts. Diffusion research has a long history in political science (Walker, 1969); however diffusion as a way to understand the policy process is largely attributed to Berry and Berry (1990, 2007).
One of the criticisms of policy diffusion is that there is no clear causal mechanisms illustrating how innovations move across states and/or countries (Gilardi, 2010; Shipan & Volden, 2008). Several recent articles have put forth specific mechanisms that can lead to policy innovation. Using antismoking policies across U.S. cities, Shipan and Volden (2008) lay out four mechanisms for policy innovation. These mechanisms include learning, economic competition, imitation, and coercion. They found that learning can occur when a similar policy is in place in other proximate cities. Second, they found that economic competition makes a city less likely to adopt antismoking policies, which are viewed to have some cost, if surrounding cities have not. Third, imitation was shown to occur by cities being more likely to adopt the policy if the nearest biggest city had a similar policy. Finally, coercion occurred as cities were less likely to adopt antismoking policies if state wide antismoking policies were in effect.
Shipan and Volden (2008) focused, in part, on the importance of learning for policy diffusion; however other work has shown that the influence of learning may be conditioned by political ideologies (Gilardi, 2010). Using a Bayesian framework, Gilardi (2010) discusses how countries learn from the experience of other countries, and how that learning can be influenced by partisan attachments. Gilardi (2010) defines learning, in a Bayesian sense, as "a process whereby policy makers change their beliefs about the effects of policies" (p. 651). Policy makers update prior beliefs based on the experiences of other countries. Gilardi (2010) uses unemployment benefits policies of 18 OECD countries and finds that right leaning governments are more likely to adopt cuts in unemployment benefits when other countries have adopted similar cuts without suffering election losses. Left leaning governments are less likely to adopt those polices if other countries experience a subsequent rise in unemployment (Gilardi, 2010).
Policy entrepreneurs have long been understood to play a role in policy development and change (Mintrom & Norman, 2009; Mintrom & Vergari, 1996); specifically they play a pivotal role within MS by taking advantage of opportunities and merging the disparate streams to enact policy change (Kingdon, 1984; Zahariadis, 2007). Mintrom and Norman define policy entrepreneurs as "highly motivated individuals or small teams [that] draw attention to policy problems, present innovative policy solutions, build coalitions of supporters, and secure legislative action" (2009, p. 649). In addition to MS, policy entrepreneurs have been thought to play a role in policy diffusion (Berry & Berry, 2007). Two recent papers examined the role that policy entrepreneurs can play in the diffusion of policies (Koski, 2010; Teodoro, 2009).
The first paper considers bureaucrats as policy entrepreneurs. Teodoro (2009) conjectures that the career mobility of bureaucrats can have an impact on the types of policy adoptions that occur across municipalities. Using a survey of municipal police chiefs and water utility managers, Teodoro (2009) is able to identify the career trajectories of bureaucrats and the hiring practices of several municipal government agencies across the U.S. Teodoro (2009) finds that agency heads that hired from within their agency are less likely to initiate policy innovation, while those that hired from outside the agency are more likely to initiate innovation. At the agency level, Teodoro finds that agencies that "routinely hire agency heads from outside the organization" (2009, p. 178) are more likely than agencies that typically hire from within to adopt new polices.
A second paper highlights the importance of policy entrepreneurs for the adoption of problems with low public salience (Koski, 2010). Using the diffusion of policies regarding the design of energy efficient building across cities (assumed to be a low-salience issue), Koski (2010) finds that policy entrepreneurs (termed knowledge brokers) played a critical role in policy adoption. With low-salience issues, knowledge brokers are needed to connect the policy idea to broader issues of public concern. In addition, policy entrepreneurs develop a "policy kernel" (a shared vocabulary of problem definitions and solutions), and they "act as linking agents between innovators and practitioners" (Koski, 2010, p. 97).
The policy diffusion model accounts for the movement of policies across governments. One of the criticisms of this model is that it lacks clear causal mechanisms that explain diffusion and adoption. The papers presented here offered several possible mechanisms. Shipan and Volden (2008) empirically tested four diffusion mechanisms including policy learning, economic competition, imitation, and coercion. Learning as a diffusion mechanism was also examined by Gilardi (2010) who found that learning occurs but is mediated by ideology. This insight has important implications for the way diffusion through learning occurs. Finally, two papers examined the role of policy entrepreneurs in policy diffusion. The first paper, characterizing bureaucrats as policy entrepreneurs, finds that bureaucratic mobility acts as a mechanism of diffusion and the second paper finds that policy entrepreneurs, as knowledge brokers, are vital to the adoption of policies related to low salience issues.
Punctuated Equilibrium is focused on two facets of policymaking; long periods of policy stasis and periods of large scale policy change. PE was initially introduced by Baumgartner and Jones (1991, 1993, 2009)8 and has subsequently developed into a theory of information processing, attention, and policy choice by governments (Jones & Baumgartner, 2005a,b; True, Jones, & Baumgartner, 2007; Workman, Jones, & Jochim, 2009). PE was originally applied to the American policy making system, however recent work has shown successful applications in other countries (Baumgartner et al., 2009; Breunig, Koski, & Mortensen, 2010; John & Jennings, 2010).9 This essay focuses on recent developments of PE as a model of policy choice based on the theory of disproportionate information processing and attention (Jones & Baumgartner, 2005a,b).10
As noted, PE has evolved beyond subsystem dynamics into a "full blown and viable model of choice for public policy" (Jones & Baumgartner, 2005a, p. 325; Mortensen, 2009). This line of research defines information as "signals" from the external environment and information processing consists of "collecting, assembling, interpreting, and prioritizing" those signals (Jones & Baumgartner, 2005b, p. 7). Processing these signals involve two aspects; selective attention and attention-driven choice (Mortensen, 2009). Selective attention, similar to bounded rationality, assumes that individuals are cognitively limited in their ability to process all available information. A similar assumption, termed institutional friction, exists for the limited ability of institutions to process information. Attention-driven choice assumes that individuals, as well as institutions, "ignore or overreact to the information signals from their surroundings" (Mortensen, 2009, p. 437) and as a result can make inefficient policy choices.
Apart from selective attention and attention-driven choice, understanding how governments process information involves understanding the related concepts of prioritization of information and the supply of information (Workman et al., 2009). Prioritization of information is important because governments are often faced with an oversupply of information. To deal with this oversupply, governments engage in both serial and parallel processing.11 The oversupply of information is argued to be a result of "pluralistic, redundant, parallel, competing, and hence 'inefficient'" processes of information gathering (Workman et al., 2009, p. 83). Workman et al. (2009) conclude that to deal with this oversupply of information Congress delegates, not only policy making authority but also information processing to the bureaucracy. This insight highlights a new and important role for the bureaucracy and its possible influence on policymaking vis-à-vis its role in information processing.
The disproportionate information processing aspects of the policy choice model has been successfully applied to questions of public budgeting (Breunig et al., 2010; Mortensen 2009; Ryu, 2009). Applying the model to public budgeting incorporates both the long-held incrementalism assumption of budgeting and the PE predictions of stability punctuated by large budgetary changes. Breunig et al. (2010) compared longitudinal budget data in Demark and the United States and found, consistent with previous research, similar patterns of aggregate budgetary changes (i.e., small, medium, and large) in both countries. Even with the aggregate similarities, the authors find diverse patterns of change and stability in more specific policy areas. In some areas the countries exhibited similar patterns of change and stability and in others there were divergent patterns. The authors conclude that further work should move beyond aggregate comparisons to account for the patterns found at the more specific policy area level.
Along similar lines, additional research finds that patterns of budget stability and punctuations can vary by type of government expenditure (Ryu, 2009). Ryu (2009), using U.S. state government budget data, finds that institutional friction and the oversupply of information can play a role in both budget stability and budget change and that these impacts vary by expenditure type. These two papers highlight the importance of understanding public budgets beyond aggregate shifts.
A final paper by Mortensen (2009) examines the attention driven choice aspect of the policy choice model. Mortensen (2009) connects public opinion with political attention (Congressional hearings), and budgetary authority. He concludes that "popular issues [measured by public opinion] tend to benefit from large increases in macro political attention but also—and perhaps even more intriguing—that unpopular issues tend to benefit from decreasing attention" (Mortensen, 2009, p. 450). This paper offers interesting insights into the importance of public opinion and political attention in determining budget authority. Political attention is assumed to be a critical component of the policy choice model and this paper empirically links political attention to a policy outcome (budgetary authority). In addition, the paper notes how public opinion, acting as information, can mediate the impacts of attention.
A large body of literature, typically using principal-agent models, argues that information asymmetries exist between Congress and the bureaucracy due to the cost to members of Congress of obtaining policy relevant information. The policy choice model argues that information is not rare and costly, but rather over abundant and as a result Congress often delegates the processing of this information to the bureaucracy (Workman et al., 2009). In other words, costs are incurred by members of Congress from the processing of information not the obtaining of information. Several empirical applications using public budgets were examined and they provided, in general, empirical support for the policy choice model. Public budgeting has been shown to exhibit the patterns of stability and change the model would expect (Breunig et al., 2010). In addition, the attention aspect of the model was shown to be significant, but conditioned by public opinion (Mortensen, 2009). Finally, as noted by Breunig et al. (2010) and Ryu (2009), the ability of the model to explain budget change at disaggregate levels and with various types of expenditures requires further development.
The social construction and policy design framework is focused on the way that attitudes regarding the target population of a policy can influence the type of policy that is created. In addition, the framework is also focused on the reciprocal—how policy can impact the way that target populations are viewed. The social construction framework was initially developed by Schneider and Ingram (1993, 1997) and later clarifications were made by Ingram, Schneider, and deLeon (2007).
A recent review of 47 applications of the social construction and policy design framework found that the framework possessed "broad utility" for both scholars and practitioners (Pierce, Schumacher, Siddiki, & Pattison, 2010, p. 20). The review examined applications by substantive policy domain and by data collection and analysis methods (Pierce et al.). Pierce et al. find that the majority of applications of the social construction and policy design framework were in the areas of health and welfare policy. Other areas included homeland security/defense, the environment, fiscal, education, immigration/race relations, and criminal justice. In terms of data collection and analysis, the majority of applications (69 percent) used qualitative methods, 20 percent were quantitative, and the remaining used mixed methods (Pierce et al.).
Recent work by Reich and Barth (2010), apply the social construction framework to policies regarding in-state tuition for undocumented college students. The authors compare in-state tuition policies across two states; Kansas and Arkansas. The states enacted different policies even though they are similar in terms of demographics and political institutions. Kansas adopted a policy that allows undocumented residents to pay in-state tuition, while similar legislation in Arkansas was not successful. Reich and Barth (2010) compared the policy deliberations that occurred in both states and found that undocumented students in Kansas where constructed as "proto-citizens", whereas the debate in Arkansas focused on questions about jurisdictional authority. They concluded that one of the key factors was the positive construction of the students in Kansas. This construction added a dimension to the debate (students, brought here as children, attempting to better their situation through education) that was able to garner enough Republican support to allow the legislation to pass. Reich and Barth (2010) provide valuable empirical and quantitative support for the importance of social constructions for policy design and adoption.
The social construction and policy design framework employs policy design as both a dependent variable and an independent variable (Ingram etal., 2007; Schneider & Sidney, 2009). Policy design consists of nine elements that include (1) problem definition and goals, (2) benefits and burdens to be distributed, (3) target population, (4) rules, (5) tools, (6) implementation structures, (7) social constructions, (8) rationales, and (9) underlying casual assumptions (Schneider & Sidney, 2009, pp. 104–105).
Some recent work has argued that policy design should be incorporated as a dependent variable in policy process theories and frameworks (James & Jorgensen, 2009; Real-Dato, 2009). James and Jorgensen (2009) argue that future work should examine the role of policy knowledge (i.e., policy analysis and policy evaluation) in determining policy design. In addition to being a dependent variable, policy design can be an independent variable in a feed-forward process. Schneider and Sidney (2009) identify four types of feed-forward effects that further work should develop. The first is the way that policy designs "create target populations," the second includes "specific rules or allocation of resources that differentially impact citizens," the third is the way that policies "embed many aspects of the rhetoric in the policy debate" specifically casual assumptions and rationales, and finally the impact of policy on pressing issues such as political participation, public cynicism, income inequality, and political rhetoric (Schneider & Sidney, 2009, p. 111).
Traditional research on policy design focused on the way in which problems were defined. The social construction and policy design framework argues scholars should also focus on the way in which target populations of a policy are defined. Building on this insight, the work by Reich and Barth (2010) found that the way that undocumented students were defined (i.e., socially constructed) played a large role in determining whether in-state tuition legislation was successful. In addition, other research brought into view considerations of the importance of policy design as both a dependent variable and an independent variable in models of the policy process. The majority of policy process research attempts to explain policy change. Scholars should be encouraged to examine issues of policy design as well.
While the limited time scale (2008 to 2010) of this essay makes identifying "trends" somewhat difficult, the following themes have emerged recently and I expect will continue to be studied by scholars of the policy process.
Recent work on the role of narratives in the policy process has offered new insight into how individuals process political and/or policy-relevant information (Jones & McBeth, 2010; McBeth et al., 2007, 2010). According to the narrative policy framework (NPF), individuals understand policy issues in terms of "stories" that include a setting or context, a plot, characters (heroes and villains), and a moral to the story (McBethJones & McBeth, 2010). The authors also argue that policy narratives need not be "relative" (i.e., context specific) but can be generalizable if "anchored" to normative beliefs. The authors suggest partisanship, ideology, and Cultural Theory (CT) as possible anchors that could guide the interpretation of policy narratives.
Jones and McBeth (2010) lay out several hypotheses, at both the micro and meso levels, that can be empirically tested. The micro level hypotheses posit predictions about individual level public opinion and the possibility of narratives to shift opinion. They hypothesize four ways that narratives can move individual opinion; (i) if a narrative alters how an individual views the world, (ii) if an individual identifies with the hero in the narrative, (iii) the degree to which the narrative is congruent with the individual's prior beliefs, and (iv) the amount of trust that the individual places in the source of the narrative. At the meso level, Jones and McBeth (2010) offer three hypotheses regarding the strategic use of narratives by groups and/or coalitions. These hypotheses are tied to classic ideas regarding conflict expansion by strategic policy actors. Jones and McBeth (2010) posit that (i) "losers" in the policy debate will use narratives to expand conflict, (ii) "winners" will employ narratives to contain conflict, and (iii) policy actors will use narratives to split opposing coalitions.
The NPF offers a way for policy scholars to empirically measure how policy relevant information is transmitted and interpreted by both policy elites and the mass public. It could possibly stand on its own as a policy process theory, or could be incorporated into existing frameworks and theories. For example, narratives can be used to explain policy-oriented learning between and across coalitions in the ACF. Narratives could also be employed by policy entrepreneurs to merge streams or hasten policy diffusion and adoption. Finally, narratives could possibly shed light on how and why information is weighted and processed by governments when making policy choices.
Policy subsystems have been the dominant level of analysis for many policy process theories and frameworks;12 particularly the ACF and PE. Recent work has brought new insights regarding subsystems. Two recent papers by Peter May and his colleagues argue that policy subsystems remain stable even after significant external disruptions (May, Sapotichne, & Workman, 2009a,b). This insight confirms long held assumptions about how subsystems bring stability to the policymaking process. In addition, recent research has begun to focus on various types of subsystems. These types include unitary, collaborative, and adversarial (Weible, 2008). The type of subsystem can have a direct bearing on the types of coalitions within that subsystem. For example, a unitary subsystem is based on a single cooperative coalition, while collaborative subsystems could have multiple coalitions, and finally an adversarial subsystem would contain multiple competing coalitions (Weible, 2008). In addition, these different subsystems are likely to use policy information and utilize policy learning in different ways (Weible & Sabatier, 2009). Finally, subsystem dynamics can help explain how policy change occurs (or doesn't) following a focusing event or crisis (Nohrstedt, 2008; Nohrstedt & Weible, 2010).
Other recent work has begun to move beyond the subsystem as the level of analysis in policymaking. PE has expanded from a model of subsystem dynamics to a system wide model of policy choice (Jones & Baumgartner, 2005a,b). In addition, recent work has argued that the ACF should move beyond the subsystem level to a more macro level policy topography (Jones & Jenkins-Smith, 2009). Other research has argued that policy problems typically encompass more than one subsystem and could be better understood as a "policy regime" (Jochim & May, 2010). Policy regimes are "governing arrangements that foster integrative actions across elements of multiple subsystems" (Jochim & May, 2010, p. 304). These regimes often emerge as a result of messy policy problems that span multiple policy areas or a crisis which can cause large scale policy disruption (May et al., 2009a, 2009b).
In large part, the bureaucracy has not been a major feature of policy process theories.13 Several recent papers have assumed a larger role for the bureaucracy and individual bureaucrats in the policymaking process. In a paper regarding information processing by governments, Workman et al. (2009) argue that Congress delegates, not only policymaking authority but also information processing to the bureaucracy. This delegation occurs as a result of the oversupply of information in the political system. Additional research regarding bureaucratic structures and information processing, argues that delegated authority and formal routines within the bureaucracy can dampen signals from political principals (Congress and the President) while centralized authority and informal procedures can amplify those signals (May, Workman, & Jones, 2008). In addition, Robinson et al. (2007) finds that bureaucratic centralization14 can make budget punctuations more likely, while punctuations are less likely in larger organizations. A final paper argues that theories of delegation to the bureaucracy should be put alongside other policy process theories or incorporated within Jones and Baumgartner's (2005a,b) policy choice model (Lavertu & Weimer, 2009). Lavertu and Weimer argue that theories of delegation contain a, "clear causal mechanism and empirically falsifiable predictions regarding the interests, information, and institutions that affect delegated policymaking," specifically delegation "explains the level of detail and the types of administrative procedures specified in statute" (2009, p. 100).
In addition to the bureaucracy, recent work has discussed the importance of individual bureaucrats in the policymaking process. Teodoro (2009) imagines bureaucrats as policy entrepreneurs and bureaucratic mobility as a causal mechanism for policy diffusion. In addition to policy entrepreneurs, recent research equates bureaucrats and public managers and argues that their role in the policy process is underdeveloped; particularly with regard to policy implementation (Hicklin & Godwin, 2009; Meier, 2009). Meier (2009) argues that there exists an overemphasis on policy design without regard to the bureaucrats that implement those policies. As Meier notes, "one of the basic facts about of implementation is that individuals, not institutions, make the majority of decisions that drive policy" (2009, p. 14).
The importance of the bureaucracy in the policymaking process has not been sufficiently considered by current policy theory. The work discussed here offer suggestions for future work that should be explored. The role of delegated information processing to the bureaucracy has important implications for policy designs and the conditions necessary for policy change. In addition, theories of delegation could be integrated into the ACF, as well as the policy choice model. It is likely that delegation patterns would vary depending on whether bureaucrats where members of the dominant coalition or the minority coalition.
Recent work regarding the policy process has largely proceeded within the established theories and frameworks. However, some work has called for integrating the various frameworks (Real-Dato, 2009; Schlager, 2007). Schlager (2007) argues that "Over the past several years, the family resemblance among the policy process theories and comparative policy models has become more pronounced, to the point where they probably belong under a single roof and that roof is the currently entitled advocacy coalition framework" (p. 317).
Along similar lines, Real-Dato (2009) argues that MS, PE, and the ACF can be joined into a single "synthetic explanatory framework." In addition, he states that the IAD could serve as theoretical "baseline" that can incorporate the synthetic framework. Using the IAD imbeds the other frameworks within a structure that accounts for the importance of institutions in the policy process, and allows for multiple levels of analysis. Real-Dato (2009) goes on to argue that the synthetic framework would incorporate three mechanisms of policy change; endogenous change, conflict expansion, and exogenous impacts. Endogenous change is change that occurs within the policy subsystem, largely as a result of policy learning. Change due to conflict expansion results from policy actors looking outside the subsystem for potential allies. Conflict expansion is based on venue-shopping by dissatisfied subsystem actors, and/or change brought about by policy actors altering the policy image of those outside of the subsystem and creating a punctuation (Real-Dato, 2009). Finally, policy change can occur as a result of exogenous impacts. Exogenous impacts can indirectly impact policy through causing an endogenous change in the subsystem or through a direct impact by causing policy change "independent of internal processes within the subsystem" (Real-Dato, 2009, p. 136).
Following Schlager (2007) and Real-Dato (2009) and focusing on merging the various theories and frameworks into a unified framework of the policy process certainly seems worth pursuing.15 The main advantage of such a unified framework is that it would allow scholars to take advantage of the cumulative knowledge of each of the frameworks.
Both before and since Sabatier's (1991) call for better theories of the policy process, multiple theories and frameworks have offered important insights into the policy process. This essay briefly outlined some recent work that has expanded those frameworks. While this essay only scratches the surface of the thriving field of policy process theory, it is my hope that one can conclude that the work examined here is of a high value and the larger field is producing interesting and exciting research. The field has generated several frameworks and theories that continue to be empirically tested and revised to sharpen our understanding of the policy process.
Matthew C. Nowlin is a doctoral candidate in political science at the University of Oklahoma. His research interests include theories of the policy process, bureaucracy and politics, and linkages between public opinion and public policy. His substantive policy area interests include nuclear energy, spent nuclear fuel management, and global climate change.
Special thanks to Hank Jenkins-Smith and two anonymous reviewers for their invaluable assistance in improving this manuscript. Any errors or omissions are the responsibility of the author.